Appearance
16.6 — X-ray, Ultrasound, CT and MRI
Wilhelm Röntgen discovered X-rays in November 1895 and, within weeks, took an image of his wife's hand. She is said to have remarked, "I have seen my death."
Within a year, X-rays were being used clinically across Europe. It remains one of the fastest translations from discovery to practice in the history of medicine.
Each imaging method answers a different question, and knowing which is which explains why a doctor orders one rather than another — and why "just do a scan" is not always the right answer.
X-ray
X-rays pass through the body and are absorbed to different degrees. What reaches the detector is a shadow.
Five densities, and reading an X-ray is largely reading the boundaries between them.
Air — black. Fat — dark grey. Soft tissue and fluid — grey, and they cannot be distinguished from each other, which is a fundamental limitation. Bone — white. Metal — bright white.
Its virtues: fast, cheap, available everywhere, portable, and low dose.
Its limitations:
No soft tissue contrast. Fluid and soft tissue look identical.
Superimposition. A three-dimensional structure projected onto a plane means everything overlaps, which is why two views at right angles are standard.
And why a lesion hidden behind the heart or below the diaphragm can be invisible on a chest film.
Insensitivity. A chest X-ray misses a substantial proportion of small lung nodules and early pneumonia. A normal chest X-ray does not exclude either.
What it is good for: chest — pneumonia, pneumothorax, heart size, pulmonary oedema; bones — fractures, arthritis; abdomen — obstruction, free gas under the diaphragm indicating perforation (Chapter 9.2); and locating foreign bodies and tubes.
Fluoroscopy — continuous X-ray, producing a moving image. Used for swallow studies, contrast studies, and to guide procedures.
And it delivers considerably more radiation than a single film, which is why exposure time is minimised.
Ultrasound
High-frequency sound, typically 2 to 15 MHz, is emitted and the returning echoes are timed.
Echoes arise at boundaries between tissues of different acoustic impedance.
Its virtues: no ionising radiation at all, real-time, portable, cheap, and it shows movement.
Which is why it is the imaging of choice in pregnancy and in children, and why it is used at the bedside in emergencies.
Its limitations, and each follows from the physics:
Air blocks it completely. So bowel gas obscures abdominal structures, and the lungs cannot be imaged conventionally — though lung ultrasound has developed by reading the artefacts that air produces.
Bone blocks it. So the adult brain cannot be imaged through the skull — but an infant's can, through the open fontanelle (Chapter 5.2), which is exactly how neonatal brain scans are done.
Depth is limited, and obesity substantially degrades the image.
And it is operator-dependent to a degree that no other modality is. The quality of the examination depends on who performs it, and a negative scan means less if the operator was inexperienced or the conditions poor.
What it is good for: pregnancy and fetal assessment; gallbladder and bile ducts — more sensitive than CT for gallstones; kidneys and bladder; liver; thyroid; testes; heart (echocardiography); blood vessels and clots; and soft tissue lumps.
Doppler measures flow by the frequency shift of the returning echo — direction and velocity, used for vessels, cardiac valves and fetal circulation.
And point-of-care ultrasound has changed emergency medicine substantially. A focused scan in a trauma patient detects free fluid in the abdomen within minutes, and it is performed by the treating clinician rather than in a radiology department.
CT
An X-ray tube rotates around the patient, taking measurements from many angles, and a computer reconstructs cross-sectional images.
Godfrey Hounsfield and Allan Cormack shared the Nobel Prize in 1979 for it, and the mathematics of reconstruction is essentially the same problem as in the transforms of Volume II, Chapter 9.
Its virtues: fast — a whole body in seconds; excellent bone detail; good soft tissue detail; and no superimposition.
Its limitations:
Radiation dose, which is the main one.
Soft tissue contrast is inferior to MRI, particularly in the brain and spinal cord.
Contrast reactions and kidney effects where iodinated contrast is used.
What it is good for: trauma — a whole-body CT in major trauma is standard and it finds injuries no examination would; acute stroke, primarily to exclude bleeding before thrombolysis; head injury; acute abdomen; pulmonary embolism, using CT pulmonary angiography; cancer staging; and lung disease.
Contrast — iodinated dye given intravenously, highlighting vessels and areas where the blood–brain barrier or vascular integrity has broken down (Chapter 11.10).
And "with contrast" versus "without" is a genuine clinical decision. A CT for a kidney stone is done without contrast, because the stone would be obscured by dye. A CT for a possible abscess is done with, because the ring of enhancement is what identifies it.
MRI
A strong magnetic field aligns the hydrogen nuclei in the body's water and fat. A radiofrequency pulse knocks them out of alignment. As they relax back, they emit a signal that depends on the local tissue environment.
Paul Lauterbur and Peter Mansfield shared the Nobel Prize in 2003.
Its virtues: no ionising radiation; outstanding soft tissue contrast — far better than CT; images in any plane; and many different sequences, each highlighting different tissue properties.
Its limitations:
Slow — 20 to 60 minutes, and the patient must stay still.
Expensive and less available.
Loud — 100 decibels or more, requiring ear protection.
Claustrophobic — a proportion of patients cannot tolerate the tunnel.
And the magnet is a genuine safety hazard. It is always on, including when no scan is running. Ferromagnetic objects are pulled in with lethal force — deaths have occurred from oxygen cylinders and other objects becoming projectiles.
Which is why screening before entry is absolute: pacemakers and implanted devices, some aneurysm clips, cochlear implants, and metal fragments in the eye — and a plain X-ray of the orbits is done first in anyone who has worked with metal, because a fragment moving inside the eye would blind them.
Most modern implants are MRI-conditional rather than absolutely contraindicated, under specified conditions.
What it is good for: brain and spinal cord — the modality of choice for both; joints, ligaments and cartilage; muscle; prostate — MRI before biopsy has substantially changed prostate cancer diagnosis (Chapter 15.1); liver lesion characterisation; bile ducts (MRCP); pelvic organs; and cancer staging in many sites.
The common sequences, briefly, because reports refer to them:
T1 — fat is bright, fluid is dark. Good anatomy. T2 — fluid is bright. Good for pathology, since most pathology involves oedema.FLAIR — T2 with the cerebrospinal fluid signal suppressed, which is why brain lesions near the ventricles become visible. Diffusion-weighted — detects restricted water movement. Positive within minutes of an ischaemic stroke, long before CT shows anything, which is why it is the standard for acute stroke imaging where available.
Nuclear medicine
The logic is inverted: the radiation source is inside the patient.
A radioactive tracer attached to a molecule that concentrates in the tissue of interest is given, and its distribution is imaged.
So nuclear imaging shows function rather than structure, which is the fundamental difference.
PET (positron emission tomography) — most often with a glucose analogue labelled with fluorine-18.
And the principle is elegant: metabolically active cells take up more glucose (Chapter 1.6). Tumours are metabolically active. So they light up.
Combined with CT as PET-CT, giving function and anatomy together, it is central to cancer staging and to assessing treatment response.
And false positives are common — infection and inflammation are also metabolically active, which is exactly why a PET-avid lesion is not automatically cancer.
Bone scan — a tracer taken up where bone turnover is increased. Metastases, fractures, infection, Paget's disease.
Thyroid scan — using the iodine-trapping mechanism (Chapter 12.3), distinguishing Graves' disease from thyroiditis.
Ventilation–perfusion scan — comparing where air goes with where blood goes. A mismatch indicates pulmonary embolism, and it is used where CT contrast is contraindicated, particularly in pregnancy.
Radiation dose
Stated with real numbers, because both complacency and excessive fear cause harm.
Doses in millisieverts, and natural background is around 2 to 3 mSv per year:
| Examination | Approximate dose | Equivalent background |
|---|---|---|
| Chest X-ray | 0.02 mSv | ~3 days |
| Dental X-ray | 0.005 mSv | ~1 day |
| Mammogram | 0.4 mSv | ~7 weeks |
| Abdominal X-ray | 0.7 mSv | ~4 months |
| CT head | 2 mSv | ~8 months |
| CT chest | 7 mSv | ~2 years |
| CT abdomen/pelvis | 8–10 mSv | ~3 years |
| PET-CT | 15–25 mSv | ~5–8 years |
| Ultrasound, MRI | 0 | — |
The risk model used for radiation protection assumes no threshold — that any dose carries some risk, proportional to the dose. This is deliberately conservative, and whether it holds at very low doses is genuinely debated.
On that model, a single abdominal CT is estimated to carry an additional lifetime fatal cancer risk in the region of 1 in 2,000.
Which is small, real, and cumulative — and it matters most in the young, whose remaining lifetime is longer and whose tissues are more radiosensitive.
Which is why:
Ultrasound and MRI are preferred in children and in pregnancy where they answer the question.
Repeated CT is scrutinised — a young person with recurrent abdominal pain can accumulate a substantial dose across several attendances, and this is a recognised problem.
And the principle is ALARA — as low as reasonably achievable.
But the risk of a missed diagnosis almost always exceeds the radiation risk when the scan is indicated. Refusing a necessary CT to avoid radiation is a bad trade, and this needs saying as clearly as the caution does.
Choosing
The practical logic:
Bone or a foreign body? X-ray.
Gallbladder, kidney, pregnancy, thyroid, testis, a soft tissue lump? Ultrasound.
Acute trauma, acute abdomen, suspected bleed in the head, suspected pulmonary embolism? CT.
Brain, spinal cord, joint, soft tissue detail, prostate? MRI.
Function, or whole-body cancer staging? Nuclear medicine.
And the questions worth asking about any proposed scan:
What are we looking for?What will change depending on the result?Is there a test without radiation that would answer it?
A scan that will not change management should generally not be done, and this is the principle behind the "choosing wisely" recommendations that exist in most specialties.
Reading a radiology report
Some conventions that make reports intelligible.
"Unremarkable" — normal.
"No acute abnormality" — nothing needing immediate action; it does not mean nothing at all.
"Clinical correlation advised" — the finding could be significant or not, depending on the symptoms. It is the radiologist saying the image alone cannot decide.
"Cannot be excluded" — the scan does not rule it out, which is different from suggesting it is present.
"Incidental finding" — something found that was not being looked for.
And incidental findings deserve emphasis, because they are common and cause disproportionate distress. They appear in a substantial minority of scans: cysts, nodules, small aneurysms, adrenal masses (Chapter 12.4).
Most are harmless. Many are given a follow-up interval, which is itself a source of anxiety, and the honest position is that follow-up exists because the natural history is uncertain rather than because the finding is worrying.
And the sensible framing: images are read in context. A finding on its own means less than the same finding in a person with matching symptoms — which is Chapter 16.4's point about pre-test probability, applied to pictures.
What Part 17 does next
The framework is complete: what disease is, how cells are injured, how the body responds, how a diagnosis is made, and what the tests mean.
From here the volume covers the diseases themselves.
And the approach changes. Parts 17 to 22 are written more briskly than the anatomy Parts — the same accuracy, told with less anatomical scaffolding, and consistently ending on what is preventable, what is treatable, and what has improved.
Because most of it has. Part 17 begins with infectious disease, where the last century has produced more lives saved than any other area of medicine.